Error with batch size 1

[israrbacha]

thanks for your work, i try your model with batch size 1 for small GPU(8GB,12GB) it doesn't wok. i check it both in ubuntu and windows;
in ubuntu it shows the error msg: floating point exception(core dumped)
in windows: it shows nothing but the execution stop.
Can you tell me the reason. BTW i check it in my friends computer too but the error is same

[HsinYingLee]

Thank you for your interest. I think the problem lies in the forward function. In current code, we need two types of input: X_encoded and X_random, which are separated by batchsize. In order to run with batch size=1, you can modify the code accordingly.

[israrbacha]

thanks for your response, another question may be irrelevant to you sorry for that. as the model is too heavy my gtx-1080ti only run 200 epochs in three days. i have an option of 2 rtx-2080ti gpu's, can you guide me how to change the code for multi-gpu's. i check the other implementation like cycleGAN,pix2pixHD but i am not successful. your code structure is different from them. thanks